[PATCH] virtio: uapi: avoid usage of libc types

[PATCH] virtio: uapi: avoid usage of libc types

[Thomas Weißschuh]

Using libc types and headers from the UAPI headers is problematic as it
introduces a dependency on a full C toolchain.

On Linux 'unsigned long' works as a replacement for 'uintptr_t' and does
not depend on libc.

Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
---
 include/uapi/linux/virtio_ring.h | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)

diff --git a/include/uapi/linux/virtio_ring.h b/include/uapi/linux/virtio_ring.h
index f8c20d3de8da..3c478582a3c2 100644
--- a/include/uapi/linux/virtio_ring.h
+++ b/include/uapi/linux/virtio_ring.h
@@ -31,9 +31,6 @@
  * SUCH DAMAGE.
  *
  * Copyright Rusty Russell IBM Corporation 2007. */
-#ifndef __KERNEL__
-#include <stdint.h>
-#endif
 #include <linux/types.h>
 #include <linux/virtio_types.h>
 
@@ -202,7 +199,7 @@ static inline void vring_init(struct vring *vr, unsigned int num, void *p,
 	vr->num = num;
 	vr->desc = p;
 	vr->avail = (struct vring_avail *)((char *)p + num * sizeof(struct vring_desc));
-	vr->used = (void *)(((uintptr_t)&vr->avail->ring[num] + sizeof(__virtio16)
+	vr->used = (void *)(((unsigned long)&vr->avail->ring[num] + sizeof(__virtio16)
 		+ align-1) & ~(align - 1));
 }
 

---
base-commit: 8f0b4cce4481fb22653697cced8d0d04027cb1e8
change-id: 20251222-uapi-virtio-d208b915ca1d

Best regards,
-- 
Thomas Weißschuh <thomas.weissschuh@linutronix.de>

Re: [PATCH] virtio: uapi: avoid usage of libc types

[Michael S. Tsirkin]

On Mon, Dec 22, 2025 at 09:00:33AM +0100, Thomas Weißschuh wrote:
> Using libc types and headers from the UAPI headers is problematic as it
> introduces a dependency on a full C toolchain.
> 
> On Linux 'unsigned long' works as a replacement for 'uintptr_t' and does
> not depend on libc.
> 
> Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>


are you fixing other uses of uintptr_t ?

> ---
>  include/uapi/linux/virtio_ring.h | 5 +----
>  1 file changed, 1 insertion(+), 4 deletions(-)
> 
> diff --git a/include/uapi/linux/virtio_ring.h b/include/uapi/linux/virtio_ring.h
> index f8c20d3de8da..3c478582a3c2 100644
> --- a/include/uapi/linux/virtio_ring.h
> +++ b/include/uapi/linux/virtio_ring.h
> @@ -31,9 +31,6 @@
>   * SUCH DAMAGE.
>   *
>   * Copyright Rusty Russell IBM Corporation 2007. */
> -#ifndef __KERNEL__
> -#include <stdint.h>
> -#endif
>  #include <linux/types.h>
>  #include <linux/virtio_types.h>
>  
> @@ -202,7 +199,7 @@ static inline void vring_init(struct vring *vr, unsigned int num, void *p,
>  	vr->num = num;
>  	vr->desc = p;
>  	vr->avail = (struct vring_avail *)((char *)p + num * sizeof(struct vring_desc));
> -	vr->used = (void *)(((uintptr_t)&vr->avail->ring[num] + sizeof(__virtio16)
> +	vr->used = (void *)(((unsigned long)&vr->avail->ring[num] + sizeof(__virtio16)
>  		+ align-1) & ~(align - 1));
>  }
>  
> 
> ---
> base-commit: 8f0b4cce4481fb22653697cced8d0d04027cb1e8
> change-id: 20251222-uapi-virtio-d208b915ca1d
> 
> Best regards,
> -- 
> Thomas Weißschuh <thomas.weissschuh@linutronix.de>

Re: [PATCH] virtio: uapi: avoid usage of libc types

[Thomas Weißschuh]

On Mon, Dec 22, 2025 at 03:39:12AM -0500, Michael S. Tsirkin wrote:
> On Mon, Dec 22, 2025 at 09:00:33AM +0100, Thomas Weißschuh wrote:
> > Using libc types and headers from the UAPI headers is problematic as it
> > introduces a dependency on a full C toolchain.
> > 
> > On Linux 'unsigned long' works as a replacement for 'uintptr_t' and does
> > not depend on libc.
> > 
> > Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
> 
> 
> are you fixing other uses of uintptr_t ?

I am focussing on UAPI headers which include libc headers.
virtio_ring.h seems to be the only such header which uses uintptr_t.
There are a few other UAPI headers which reference uintptr_t, but only
in comments or macros.

uintptr_t in regular kernel code is out of scope for me.
While it is iffy, it doesn't actually hurt.


Thomas

Re: [PATCH] virtio: uapi: avoid usage of libc types

[Michael S. Tsirkin]

On Mon, Dec 22, 2025 at 11:22:16AM +0100, Thomas Weißschuh wrote:
> On Mon, Dec 22, 2025 at 03:39:12AM -0500, Michael S. Tsirkin wrote:
> > On Mon, Dec 22, 2025 at 09:00:33AM +0100, Thomas Weißschuh wrote:
> > > Using libc types and headers from the UAPI headers is problematic as it
> > > introduces a dependency on a full C toolchain.
> > > 
> > > On Linux 'unsigned long' works as a replacement for 'uintptr_t' and does
> > > not depend on libc.
> > > 
> > > Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
> > 
> > 
> > are you fixing other uses of uintptr_t ?
> 
> I am focussing on UAPI headers which include libc headers.
> virtio_ring.h seems to be the only such header which uses uintptr_t.
> There are a few other UAPI headers which reference uintptr_t, but only
> in comments or macros.
> 
> uintptr_t in regular kernel code is out of scope for me.
> While it is iffy, it doesn't actually hurt.
> 
> 
> Thomas

And other uses of stdint.h?

Re: [PATCH] virtio: uapi: avoid usage of libc types

[Thomas Weißschuh]

On Mon, Dec 22, 2025 at 05:28:30AM -0500, Michael S. Tsirkin wrote:
> On Mon, Dec 22, 2025 at 11:22:16AM +0100, Thomas Weißschuh wrote:
> > On Mon, Dec 22, 2025 at 03:39:12AM -0500, Michael S. Tsirkin wrote:
> > > On Mon, Dec 22, 2025 at 09:00:33AM +0100, Thomas Weißschuh wrote:
> > > > Using libc types and headers from the UAPI headers is problematic as it
> > > > introduces a dependency on a full C toolchain.
> > > > 
> > > > On Linux 'unsigned long' works as a replacement for 'uintptr_t' and does
> > > > not depend on libc.
> > > > 
> > > > Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
> > > 
> > > 
> > > are you fixing other uses of uintptr_t ?
> > 
> > I am focussing on UAPI headers which include libc headers.
> > virtio_ring.h seems to be the only such header which uses uintptr_t.
> > There are a few other UAPI headers which reference uintptr_t, but only
> > in comments or macros.
> > 
> > uintptr_t in regular kernel code is out of scope for me.
> > While it is iffy, it doesn't actually hurt.
> > 
> > 
> > Thomas
> 
> And other uses of stdint.h?

If they are in UAPI headers I will look at them.
I also have a series prepared to prevent new users from creeping in.

Re: [PATCH] virtio: uapi: avoid usage of libc types

[Michael S. Tsirkin]

On Mon, Dec 22, 2025 at 11:43:52AM +0100, Thomas Weißschuh wrote:
> On Mon, Dec 22, 2025 at 05:28:30AM -0500, Michael S. Tsirkin wrote:
> > On Mon, Dec 22, 2025 at 11:22:16AM +0100, Thomas Weißschuh wrote:
> > > On Mon, Dec 22, 2025 at 03:39:12AM -0500, Michael S. Tsirkin wrote:
> > > > On Mon, Dec 22, 2025 at 09:00:33AM +0100, Thomas Weißschuh wrote:
> > > > > Using libc types and headers from the UAPI headers is problematic as it
> > > > > introduces a dependency on a full C toolchain.
> > > > > 
> > > > > On Linux 'unsigned long' works as a replacement for 'uintptr_t' and does
> > > > > not depend on libc.
> > > > > 
> > > > > Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
> > > > 
> > > > 
> > > > are you fixing other uses of uintptr_t ?
> > > 
> > > I am focussing on UAPI headers which include libc headers.
> > > virtio_ring.h seems to be the only such header which uses uintptr_t.
> > > There are a few other UAPI headers which reference uintptr_t, but only
> > > in comments or macros.
> > > 
> > > uintptr_t in regular kernel code is out of scope for me.
> > > While it is iffy, it doesn't actually hurt.
> > > 
> > > 
> > > Thomas
> > 
> > And other uses of stdint.h?
> 
> If they are in UAPI headers I will look at them.
> I also have a series prepared to prevent new users from creeping in.

I don't mind picking this but I'm just curious what's driving all this
effort.

-- 
MST

Re: [PATCH] virtio: uapi: avoid usage of libc types

[Thomas Weißschuh]

On Mon, Dec 22, 2025 at 05:18:26PM -0500, Michael S. Tsirkin wrote:
> On Mon, Dec 22, 2025 at 11:43:52AM +0100, Thomas Weißschuh wrote:
> > On Mon, Dec 22, 2025 at 05:28:30AM -0500, Michael S. Tsirkin wrote:
> > > On Mon, Dec 22, 2025 at 11:22:16AM +0100, Thomas Weißschuh wrote:
> > > > On Mon, Dec 22, 2025 at 03:39:12AM -0500, Michael S. Tsirkin wrote:
> > > > > On Mon, Dec 22, 2025 at 09:00:33AM +0100, Thomas Weißschuh wrote:
> > > > > > Using libc types and headers from the UAPI headers is problematic as it
> > > > > > introduces a dependency on a full C toolchain.
> > > > > > 
> > > > > > On Linux 'unsigned long' works as a replacement for 'uintptr_t' and does
> > > > > > not depend on libc.
> > > > > > 
> > > > > > Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
> > > > > 
> > > > > 
> > > > > are you fixing other uses of uintptr_t ?
> > > > 
> > > > I am focussing on UAPI headers which include libc headers.
> > > > virtio_ring.h seems to be the only such header which uses uintptr_t.
> > > > There are a few other UAPI headers which reference uintptr_t, but only
> > > > in comments or macros.
> > > > 
> > > > uintptr_t in regular kernel code is out of scope for me.
> > > > While it is iffy, it doesn't actually hurt.
> > > > 
> > > > 
> > > > Thomas
> > > 
> > > And other uses of stdint.h?
> > 
> > If they are in UAPI headers I will look at them.
> > I also have a series prepared to prevent new users from creeping in.
> 
> I don't mind picking this but I'm just curious what's driving all this
> effort.

It started with a recent change to an UAPI header which introduced a dependency
on libc that broke some things I was doing. See the fix and background in [0].
So I looked into catching such issues earlier, which resulted in [1].
Fortunately there are not many problematic headers [2] in general, so I am now
also fixing those, so the new check can validate more headers and prevent more
regressions.

[0] https://lore.kernel.org/lkml/20251203-uapi-fcntl-v1-1-490c67bf3425@linutronix.de/
[1] https://lore.kernel.org/lkml/20251223-uapi-nostdinc-v1-0-d91545d794f7@linutronix.de/
[2] https://lore.kernel.org/lkml/20251223-uapi-nostdinc-v1-1-d91545d794f7@linutronix.de/


Thomas

Re: [PATCH] virtio: uapi: avoid usage of libc types

[Arnd Bergmann]

On Mon, Dec 22, 2025, at 09:00, Thomas Weißschuh wrote:
> Using libc types and headers from the UAPI headers is problematic as it
> introduces a dependency on a full C toolchain.
>
> On Linux 'unsigned long' works as a replacement for 'uintptr_t' and does
> not depend on libc.
>
> Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>

Acked-by: Arnd Bergmann <arnd@arndb.de>